Prevention of DNA Rereplication Through a Meiotic Recombination Checkpoint Response

In the budding yeast Saccharomyces cerevisiae, unnatural stabilization of the cyclin-dependent kinase inhibitor Sic1 during meiosis can trigger extra rounds of DNA replication. When programmed DNA double-strand breaks (DSBs) are generated but not repaired due to absence of DMC1, a pathway involving the checkpoint gene RAD17 prevents this DNA rereplication. Further genetic analysis has now revealed that prevention of DNA rereplication also requires MEC1, which encodes a protein kinase that serves as a central checkpoint regulator in several pathways including the meiotic recombination checkpoint response. Downstream of MEC1, MEK1 is required through its function to inhibit repair between sister chromatids. By contrast, meiotic recombination checkpoint effectors that regulate gene expression and cyclin-dependent kinase activity are not necessary. Phosphorylation of histone H2A, which is catalyzed by Mec1 and the related Tel1 protein kinase in response to DSBs, and can help coordinate activation of the Rad53 checkpoint protein kinase in the mitotic cell cycle, is required for the full checkpoint response. Phosphorylation sites that are targeted by Rad53 in a mitotic S phase checkpoint response are also involved, based on the behavior of cells containing mutations in the DBF4 and SLD3 DNA replication genes. However, RAD53 does not appear to be required, nor does RAD9, which encodes a mediator of Rad53, consistent with their lack of function in the recombination checkpoint pathway that prevents meiotic progression. While this response is similar to a checkpoint mechanism that inhibits initiation of DNA replication in the mitotic cell cycle, the evidence points to a new variation on DNA replication control.

ABSTRACT In the budding yeast Saccharomyces cerevisiae, unnatural stabilization of the cyclin-dependent kinase inhibitor Sic1 during meiosis can trigger extra rounds of DNA replication. When programmed DNA double-strand breaks (DSBs) are generated but not repaired due to absence of DMC1, a pathway involving the checkpoint gene RAD17 prevents this DNA rereplication. Further genetic analysis has now revealed that prevention of DNA rereplication also requires MEC1, which encodes a protein kinase that serves as a central checkpoint regulator in several pathways including the meiotic recombination checkpoint response. Downstream of MEC1, MEK1 is required through its function to inhibit repair between sister chromatids. By contrast, meiotic recombination checkpoint effectors that regulate gene expression and cyclin-dependent kinase activity are not necessary. Phosphorylation of histone H2A, which is catalyzed by Mec1 and the related Tel1 protein kinase in response to DSBs, and can help coordinate activation of the Rad53 checkpoint protein kinase in the mitotic cell cycle, is required for the full checkpoint response. Phosphorylation sites that are targeted by Rad53 in a mitotic S phase checkpoint response are also involved, based on the behavior of cells containing mutations in the DBF4 and SLD3 DNA replication genes. However, RAD53 does not appear to be required, nor does RAD9, which encodes a mediator of Rad53, consistent with their lack of function in the recombination checkpoint pathway that prevents meiotic progression. While this response is similar to a checkpoint mechanism that inhibits initiation of DNA replication in the mitotic cell cycle, the evidence points to a new variation on DNA replication control.
In the absence of the meiosis-specific DNA recombinase Dmc1, strand invasion cannot proceed and extensive DNA resection results, leading to activation of a meiotic recombination checkpoint response that prevents exit from the pachytene stage of prophase I (Bishop et al. 1992;Xu et al. 1997).
As might be expected, the meiotic recombination checkpoint pathway as defined by deletion of the DMC1 gene (dmc1D) shares many proteins with DNA damage checkpoint pathways that operate during the mitotic cell cycle (Lydall et al. 1996). Examples include the apical protein kinase Mec1, and its associated protein Ddc2, which are orthologs of human ATM-and Rad3-related protein kinase (ATR), and ATR-interacting protein (ATRIP), respectively, and the PCNA-like Ddc1-Mec3-Rad17 ("9-1-1") complex, which facilitates Mec1 function, and also has a human counterpart (Weinert et al. 1994;Lydall et al. 1996;Paciotti et al. 2000;Hong and Roeder 2002;Zou and Elledge 2003;Navadgi-Patil and Burgers 2011;Refolio et al. 2011). In addition, the Dot1 methyltransferase is involved in both (San-Segundo and Roeder 2000; Giannattasio et al. 2005;Wysocki et al. 2005). By contrast, the protein kinase Rad53, an ortholog of human CHK2, and its mediator Rad9, similar in some respects to human mediators such as BRCA1 and 53BP1, function downstream of Mec1 in various cell cycle DNA damage checkpoints (Allen et al. 1994;Weinert et al. 1994;Sun et al. 1996;Gilbert et al. 2001;Stracker et al. 2009), but are not involved in the meiotic recombination checkpoint (Bishop et al. 1992;Lydall et al. 1996;Bailis and Roeder 2000). While Rad53 and Rad9 have been implicated in certain meiotic checkpoints, including the response to hydroxyurea (HU) (S-phase) (Blitzblau and Hochwagen 2013), and to unprogrammed DNA damage (Weber and Byers 1992;Cartagena-Lirola et al. 2008), their absence in the recombination checkpoint can be explained by the existence of meiosis-specific proteins that operate specifically in the context of recombination intermediate structures (Hollingsworth and Ponte 1997;Xu et al. 1997;Bailis and Roeder 2000). These include Hop1, Red1, and Mek1, each of which is a component of the sister chromatid-derived axial elements that form during meiosis, and are critical for proper meiotic recombination. Hop1 and Red1 are structural in nature (Hollingsworth et al. 1990;Smith and Roeder 1997), whereas Mek1 is a protein kinase with sequence similarity to Rad53 (Rockmill and Roeder 1991;Leem and Ogawa 1992;Bailis and Roeder 2000). All three proteins help to enforce the proper bias of interhomolog recombination during unperturbed meiosis, thereby promoting faithful chromosome segregation (Hollingsworth and Byers 1989;Rockmill and Roeder 1991;Schwacha and Kleckner 1997;Thompson and Stahl 1999;Kim et al. 2010;Wu et al. 2010). In the context of the meiotic recombination checkpoint activated by deletion of DMC1, Red1 associates with the 9-1-1 complex to help activate Mec1 (Eichinger and Jentsch 2010), leading to Mec1-catalyzed Hop1 phosphorylation required for Mek1 activation (Carballo et al. 2008). Mek1 activity in turn prevents excessive recombination between sister chromatids, and thereby maintains the checkpoint signal (Wan et al. 2004;Niu et al. 2005). Studies using mutants with defects in intersister DSB repair, or those in which the DSBs that are generated cannot be efficiently repaired, have shown that Mek1 also serves to prevent meiotic progression (Xu et al. 1997;Cartagena-Lirola et al. 2008;Wu et al. 2010).
Ultimately, checkpoint-mediated prevention of pachytene exit and progression through the meiotic divisions is implemented in part through regulation of the NDT80 gene and its protein product, which is a meiosis-specific transcription factor required for proper expression of many "middle" sporulation genes (Chu and Herskowitz 1998;Hepworth et al. 1998;Lindgren et al. 2000;Tung et al. 2000;Pak and Segall 2002;Shubassi et al. 2003). These include CDC5, whose polo-like protein kinase product is required for pachytene exit, and also upregulates Ndt80 activation in a feedback loop (Sourirajan and Lichten 2008;Acosta et al. 2011), and CLB1, which encodes a B-type cyclin that is required for progression through meiosis I (Chu and Herskowitz 1998;Carlile and Amon 2008). Another target of the meiotic recombination checkpoint is the Swe1 protein kinase, which is activated to catalyze inhibitory phosphorylation of Cdk1 at tyrosine 19 (Leu and Roeder 1999). Early work in mitotic cells indicated that Swe1-catalyzed Cdk1 phosphorylation regulates the morphogenesis checkpoint (Lew and Reed 1995). However, it is now known that Swe1 is also a component of one of three Mec1-dependent mechanisms that operate in the S phase checkpoint to prevent cell cycle progression into mitosis (Palou et al. 2015).
In our previous studies, we found that deletion of DMC1 blocks the DNA rereplication induced by Sic1 stabilization (Sawarynski et al. 2009). In this report, we describe our further genetic investigation into constituents of the meiotic recombination checkpoint as they pertain to Sic1-induced DNA rereplication. We found that certain upstream components, including MEC1, were required to prevent DNA rereplication. However, we did not find evidence that particular downstream effectors that regulate meiotic progression were involved. We further examined Figure 1 MEC1 is required for dmc1D-dependent inhibition of SIC1ΔP HA -induced DNA rereplication. Strains were treated to enter the meiotic program in a synchronous fashion. At the indicated time points, samples were analyzed for Sic1DP HA (HA) and the tubulin control (tub) by western blotting (upper panels), and for DNA content by flow cytometry (lower histograms). For this figure, and others that follow, the 4C designation indicates the population of cells that have undergone one round of DNA replication; cells with rereplicated DNA appear to the right of this position, with peaks of cells and approximate DNA contents .4C indicated. Conclusions for this experiment are provided as schematics to provide an example of pathway analysis.
processes that operate to prevent DNA replication in the mitotic cell cycle, and found overlap with respect to specific phosphorylation events, including those that are important for blocking late DNA replication origin firing in an S phase checkpoint response. Interestingly, these data suggest a pathway in which the effectors are phosphorylated through a Rad53-independent mechanism.

Strains
Yeast strains used in this study are listed in Table 1. Construction of the HOP1pr-SIC1ΔP HA module, and its integration into the genome, were described previously (Sawarynski et al. 2009). In most cases, deletion mutations were generated in haploids by homology-directed sitespecific replacement with selectable markers (Baudin et al. 1993). These markers were PCR-amplified from either genomic DNA of a deletion set mutant (Winzeler et al. 1999) (GE Dharmacon), or from a plasmid (Brachmann et al. 1998). Certain mutant progenitor strains in the W303 background were generously provided by other investigators: SKY2939 (h2a-S129A) (Downs et al. 2004) by Stephen Kron (University of Chicago), YFL234 (dot1D) (Giannattasio et al. 2005) by Marco Muzi-Falconi (Università degli Studi di Milano), U960 (rad53D sml1-1) (Zhao et al. 1998) by Stephen Elledge (Harvard University), and Y2359 and Y2573 (dbf4-4A, sld3-38A, and mcm5-bob1) (Zegerman and Diffley 2010) by Philip Zegerman (The Gurdon Institute, UK) and John Diffley (The Francis Crick Institute, UK). These mutations were then introduced into our cell system through crossing. Strain construction generally involved introduction of mutations into MATa cells and into MATa cells with the HOP1pr-SIC1ΔP HA module either present or subsequently added, followed by mating of the two cell types. (Note that the shorthand designation of SIC1ΔP HA used for diploids in the text and figures indicates the presence of a single copy of the HOP1pr-SIC1ΔP HA element, while other mutant allele designations indicate alteration of both gene copies.) All deletion mutations generated for this study were verified by PCR, and deletion of SWE1 was further confirmed by western blotting using antibody kindly provided by Doug Kellogg (University of California, Santa Cruz) (Sreenivasan and Kellogg 1999). DNA sequencing was used to validate the presence of certain point mutations in our strains. Epitope tagging of Sic1 (SIC1 13MYC ) was performed as described (Longtine et al. 1998) in MATa and MATa cells, which were then mated to generate the diploid. An additional manipulation included 5-fluoro-orotic acid-mediated counter-selection (Boeke et al. 1984) to isolate a rad53D sml1-1 diploid from a strain containing HOP1pr-SIC1DP HA .

Figure 2
Factors that enforce interhomolog bias and prevent meiotic progression are required for dmc1D-dependent inhibition of SIC1ΔP HAinduced DNA rereplication. Cells were treated to enter meiosis and analyzed for protein levels by western blotting, and for DNA content by flow cytometry. (A, B), the effect of red1Δ and hop1Δ, respectively; (C), the effect of mek1Δ, as well as rad54Δ. Western blotting analysis included Sic1DP HA (HA), tubulin (tub) and phosphorylated H2A (g-H2A).

Cell culture
All yeast incubations were conducted at 30°. Meiosis was induced by starvation based on an established procedure for synchronous sporulation (Padmore et al. 1991). In this method, yeast cells were first grown on solid [2% (w/v) agar] YPG medium [1% (w/v) yeast extract, 2% (w/v) peptone, 3% (v/v) glycerol], or, alternatively, on solid YPD medium [1% (w/v) yeast extract, 2% (w/v) peptone, 2% (w/v) dextrose] for 3-4 d, and single colonies were used to inoculate YPD liquid cultures. The overnight YPD cultures were then used to inoculate YPA [1% (w/v) yeast extract, 2% (w/v) peptone, 2% (w/v) potassium acetate] at an OD 600 of 0.2. Cells were incubated overnight, typically for 16 hr, and then resuspended at equivalent cell densities (based on OD 600 values) for strains within an experiment in sporulation medium consisting of 0.3% (w/v) potassium acetate and 0.02% (w/v) raffinose supplemented with leucine, arginine, and histidine each at 250 mM, tryptophan at 100 mM, and uracil at 50 mM. Cells were returned to incubation at time 0, and aliquots were harvested at indicated time points for flow cytometry and protein analyses (see below). For most experiments, control strains SIC1ΔP HA and SIC1ΔP HA dmc1Δ were included. Each experimental strain was analyzed at least twice in independent experiments, and in many cases more than twice (see Supplemental Material, Figure S1).
To conduct a synchronized mitotic cell cycle timecourse, MATa cells grown to saturation in YPD were brought to an OD 600 of 0.2 and incubated for 2 hr. The yeast mating pheromone a-factor (Zymo Research) was then added to a final concentration of 2.5 mM, and cells were incubated for an additional 2 hr to achieve G1 arrest. The cells were then washed with sterile water to remove a-factor, resuspended in fresh YPD, and further incubated. Aliquots were taken at 15-min intervals for flow cytometry and western blotting analyses (see below). For examining the response to inhibition of DNA replication, cells were arrested with a-factor as described above and then released into 0.8· YPD containing 0.2 M HU (MP Biomedicals).

DNA content
Cells were harvested by centrifugation, resuspended in 70% ethanol and stored at 4°. Aliquots of the fixed cells were washed once with 50 mM Tris-HCl, pH 7.5, resuspended in 1 ml of the same buffer, and then treated with 250 mg RNase A for 1 hr at 37°, followed by 250 mg proteinase K for 1 hr at 37°. Digested samples were incubated with 10· SYBR Green I (Molecular Probes) at 4°overnight, sonicated briefly, and analyzed with a FACSCanto II flow cytometer (BD Biosciences) (or, in one experiment, a BD LSR II flow cytometer) (Microscopy, Imaging, and Cytometry Resources Core at Wayne State University School of Medicine). DNA content histograms were generated using WinMDI freeware. DNA rereplication was quantified by using the gating function in WinMDI to determine the number of events (out of 20,000) that were recorded with . 4C DNA content. Gating was established based on the 4C DNA peak, and was held constant within each individual experiment.

Protein
Cells were harvested by centrifugation and stored at 270°. For most experiments, denatured whole-cell extracts were prepared based on an alkaline extraction method (Kushnirov 2000). In the case of Rad53 analysis, a tricholoroacetic acid bead beating method was used, as described (Foiani et al. 1994). Resulting samples were subjected to SDSpolyacrylamide electrophoresis. For western blotting, the separated proteins were transferred to nitrocellulose (GE Healthcare). Primary antibodies included rat anti-a-tubulin (Serotec), mouse anti-hemagglutinin (Covance), rabbit anti-yeast g-H2A (generously provided by Christophe Redon and William Bonner, National Cancer Institute) (Nakamura et al. 2006), mouse anti-myc (Santa Cruz), and rabbit anti-Rad53 (Abcam). Signals were generated with IRDye 800-conjugated goat antirat (Rockland), Alexa Fluor 680 goat anti-rabbit (Invitrogen), or Alexa Fluor 680 goat anti-mouse (Invitrogen) secondary antibodies. Reactive bands were visualized with an Odyssey infrared imaging system (Li-Cor). For analysis of Rad53 activity, separated proteins were transferred to PVDF (Millipore), and Rad53 autophosphorylation was analyzed in situ as described (Foiani et al. 1994). In the figures, line borders indicate cropping, and vertically contiguous panels indicate data originating from the same blot or membrane.

Data availability
The authors state that all data necessary for confirming the conclusions presented in the article are represented fully within the article.

RESULTS
We have developed a system in which cells undergoing meiosis experience at least one extra round of DNA replication (Sawarynski et al. 2009). This phenotype occurs upon deregulation of CDK activity through early meiosis-specific expression (via the HOP1 promoter) of SIC1ΔP HA , encoding a Sic1 HA variant lacking critical CDK phosphorylation sites that are necessary for its destruction (Verma et al. 1997). DNA rereplication does not occur in this system when the Sic1 HA phosphorylation sites are not altered, as this version of the protein is subject to post-translational modification and degradation. Interestingly, deletion of DMC1 (dmc1D) activates a response that prevents the DNA rereplication phenotype normally observed in our specially engineered cells. We have examined this pathway genetically, as described below. The extent of DNA rereplication in the various strains that we describe in this study was quantified from flow cytometry data, and is presented as a compilation in Figure S1.

MEC1
Previously, we showed that RAD17, which encodes a 9-1-1 member, is required to suppress dmc1D-dependent inhibition of DNA rereplication, suggesting that a branch of the meiotic recombination checkpoint could affect DNA replication (Sawarynski et al. 2009). Given that Rad17 is intimately associated with Mec1, we suspected that this central protein kinase was also involved. To test this hypothesis, we generated a mec1 null mutant in our strain background. Because MEC1 is essential for viability (Kato and Ogawa 1994), it was necessary to delete SML1 as well to suppress the lethality resulting from MEC1 loss (Zhao et al. 1998). As shown in Figure 1, DNA rereplication was observed in SIC1ΔP HA dmc1Δ mec1Δ sml1Δ cells. This phenotype was not due to sml1Δ, as SIC1ΔP HA dmc1Δ sml1Δ cells did not exhibit DNA rereplication (see Figure S1). Therefore, MEC1 was involved in the prevention of DNA rereplication in SIC1ΔP HA dmc1Δ cells. We observed less robust DNA rereplication in SIC1ΔP HA dmc1Δ mec1Δ sml1Δ cells than in SIC1ΔP HA cells; a portion of this effect may have been due to the absence of MEC1 and SML1 as revealed by examination of SIC1ΔP HA mec1Δ sml1Δ cells (Figure 1 and Figure S1).

MEK1 and associated genes
We further examined genes that operate downstream of MEC1 in the established response that prevents both intersister repair and meiotic progression, including those that encode the axial proteins Mek1, Red1, and Hop1. As in the case of mec1Δ, deletion of any one of these genes restored DNA rereplication in SIC1ΔP HA dmc1D cells ( Figure  2), indicating that they were required for prevention of DNA rereplication in response to the accumulation of unrepaired DSBs. In these cases, DNA rereplication was robust, as exhibited by the generation of cells with high DNA content in some cases reaching 16C (Figure 2 and Figure S1).
It has been shown that dmc1Δ mek1Δ cells that are also rad54Δ, and therefore incapable of completing intersister repair (Arbel et al. 1999) progress through meiosis, albeit with slower kinetics than wild-type or dmc1D mek1D cells (Cartagena-Lirola et al. 2008;Chuang et al. 2012); this phenotype illustrates the checkpoint function of MEK1. To determine whether the MEK1 function to suppress intersister repair, or to prevent meiotic progression, was at play in our specialized case, we examined SIC1ΔP HA dmc1Δ mek1Δ rad54Δ cells. We found that the recovery of DNA rereplication observed in SIC1ΔP HA dmc1D mek1D cells was not observed with the addition of rad54Δ ( Figure 2C). In each strain, we observed an increase in phosphorylated histone H2A (g-H2A), which is generated through Mec1 (and related Tel1) catalysis in response to DSB formation, and leads to extensive regions of chromatin containing g-H2A on either side of the DSB (Shroff et al. 2004). These data suggest a persistence of DSBs in our cells throughout the time course, regardless of DMC1 or RAD54 status. We further demonstrated that RAD54 itself was not required for the DNA rereplication phenotype ( Figure S2). These data indicate that MEK1 inhibited DNA rereplication by preventing intersister repair and maintaining the DSBinduced signal rather than by influencing DNA replication itself. Figure 4 The CDK inhibitor Sic1 does not accumulate in SIC1ΔP HA dmc1D cells. The indicated cells containing Sic1 13MYC were treated to enter meiosis, and then analyzed at several time points for Sic1DP HA (HA), Sic1 13MYC (MYC) and tubulin (tub) by western blotting, and for DNA content by flow cytometry.
Studies have indicated that the AAA+-type ATPase Pch2 suppresses intersister repair to some extent, and helps to prevent meiotic progression when unrepaired DSBs accumulate (San-Segundo and Roeder 1999; Ho and Burgess 2011;Zanders et al. 2011;Chen et al. 2014). We found that deletion of PCH2 relieved the dmc1Δ-dependent inhibition of DNA rereplication in our SIC1ΔP HA system ( Figure S3), indicating that PCH2 aided in preventing DNA rereplication. In this case, as in certain other mutants analyzed (see below), few cells with DNA content .8C were observed. Extensive DNA rereplication was detected with SIC1ΔP HA pch2Δ cells ( Figure S1 and Figure S3), indicating that PCH2 itself was not required for the DNA rereplication phenotype in SIC1ΔP HA cells.

CDK
We considered the possibility that the checkpoint response might prevent CDK activation to inhibit DNA rereplication. One mechanism by which the meiotic recombination checkpoint prevents meiotic progression is through Sum1, a key transcription factor that represses expression of middle sporulation genes normally induced by the transcription factor Ndt80 (Lindgren et al. 2000;Pak and Segall 2002;Winter 2012). Included among these Ndt80-induced genes are those that encode the B-type cyclins Clb1, -3, -4, -5, and -6 (Chu and Herskowitz 1998). We found that DNA rereplication occurred in SIC1ΔP HA sum1D cells, although with reduced efficiency for cells with DNA content .8C, but not in SIC1ΔP HA dmc1D sum1D cells ( Figure 3A and Figure S1), indicating that SUM1 was not involved in this checkpoint.
We also examined SWE1, whose product becomes activated in the meiotic recombination checkpoint response to catalyze inhibitory phosphorylation of Cdk1 at tyrosine 19 (Leu and Roeder 1999). We found that deletion of SWE1, like that of SUM1, did not prevent DNA rereplication in SIC1ΔP HA cells, nor did it reverse the dmc1D-dependent inhibition of DNA rereplication (Figure 3, B and C), suggesting that SWE1 was not required for prevention of DNA rereplication. It is noted that because Ndt80 levels and activity are downregulated by the meiotic recombination checkpoint response, thereby lowering B-type cyclin availability and CDK activity, we might not expect to see much of an effect by simply deleting SWE1 in our cells. However, dmc1Δ swe1Δ cells do progress into MI, although with delayed kinetics relative to wild type cells (Leu and Roeder 1999). During the course of these studies, we also examined a swe1D mutant without the SIC1ΔP HA allele. This experiment was prompted by the report that swe1D cells rereplicate their DNA during meiosis, a phenotype that is different from ours in that multispore asci are formed (Rice et al. 2005). We did not observe these phenotypes in our swe1D cells, perhaps due to differences in strain types or culture conditions ( Figure 3B and data not shown).
To investigate another CDK regulator, Sic1, we generated strains in which Sic1 was tagged with MYC epitope repeats. In this way, we could distinguish endogenous Sic1 from the induced version lacking CDKtargeted phosphorylation sites, which is tagged with the HA epitope. We first examined a haploid strain during the cell cycle to ensure that our Sic1 13MYC protein behaved properly. As expected, we found that Sic1 disappeared almost completely as cells progressed from G1 arrest, and then reappeared after S phase was completed ( Figure S4). We next examined diploid strains in meiosis (Figure 4). Synchrony is difficult to achieve in meiosis, and we did not observe the sharp reduction and reappearance of Sic1 in our wild-type cells. However, we did see a decline in Sic1 as cells progressed through the time course, which may have reflected a real decrease in cellular Sic1 steady state levels. As shown, we observed nearly identical Sic1 13MYC profiles in SIC1ΔP HA and SIC1ΔP HA dmc1D cells. We conclude that a decrease in CDK activity through stabilization of endogenous Sic1 was not likely to be responsible for dmc1D-dependent inhibition of DNA rereplication. g-H2A and DOT1 Mec1-and Tel1-catalyzed H2A phosphorylation at serine 129, which generates g-H2A, is thought to be an important protective mechanism that promotes DSB repair to prevent genomic alterations (Downs et al. 2000;Redon et al. 2003), and functions in the G1 DNA damage checkpoint during the mitotic cell cycle (Javaheri et al. 2006;Hammet et al. 2007). Because MEC1 was required in the meiotic recombination checkpoint response that prevents DNA rereplication, we suspected that one of its major targets would be involved as well. We found that g-H2A was generated in H2A (HTA1 and HTA2) cells regardless of DMC1 status (see Figure 2C). We generated a SIC1DP HA strain with HTA1 and HTA2 both mutated (h2a-S129A) so that the two H2A subunits could not be phosphorylated at serine 129, and confirmed through western blotting that these cells were devoid of g-H2A (see Figure S2 and Figure S5). Importantly, absence of g-H2A led to DNA rereplication in SIC1ΔP HA dmc1D cells; while not as extensive as in SIC1ΔP HA cells with regard to total number of cells exhibiting .4C DNA content and those reaching .8C, the DNA rereplication was obvious, and observed consistently ( Figure 5 and Figure S1). SIC1ΔP HA h2a-S129A cells rereplicated their DNA with clear evidence of cells containing .8C DNA content ( Figure S2).
We also examined the phenotype resulting from deletion of DOT1, which encodes a histone methyltransferase required for the meiotic recombination checkpoint response (San-Segundo and Roeder 2000). Dot1 catalyzes methylation of histone H3 at lysine K79 (H3meK79) (Lacoste et al. 2002;Ng et al. 2002;van Leeuwen et al. 2002), and, like g-H2A, is important for the G1 DNA damage checkpoint in the mitotic cell cycle (Giannattasio et al. 2005;Wysocki et al. 2005). Similar to the case with h2a-S129A, we observed a modest degree of DNA rereplication in SIC1ΔP HA dmc1D dot1Δ cells, and combination of the dot1Δ and h2a-S129A mutations did not enhance this effect ( Figure 5 and Figure S1). These data suggest that g-H2A and Dot1 operated in the same pathway in preventing DNA rereplication in SIC1ΔP HA dmc1D cells. Figure 6 Rad53 is not activated by deletion of DMC1. (A) The indicated strains were treated to enter meiosis and then analyzed for: Sic1DP HA (HA) and tubulin (tub) by western blotting, and for DNA content by flow cytometry. (B) Samples from the same time course shown in (A) were used to assess Rad53 activation through western blotting [Rad53 and tubulin (tub) top panel], and Rad53 autophosphorylation in situ ( 32 P-Rad53, middle panel). Total protein loading for the autophosphorylation assay was determined by Ponceau S staining (bottom panel). Control samples included vegetative SIC1DP HA dmc1D (RAD53) and rad53D sml1-1 (rad53D) diploids exposed to HU for the times indicated.

RAD53
While Rad53 is not involved in the meiotic recombination checkpoint per se, it can be activated by genotoxic stress during meiosis (Weber and Byers 1992;Cartagena-Lirola et al. 2008;Blitzblau and Hochwagen 2013). Therefore, we elected to determine whether or not RAD53 was involved in the checkpoint that prevents DNA rereplication in our system. (As in the case of mec1D, rad53D cells are viable when SML1 is also defective (Zhao et al. 1998)). We were surprised to find that DNA rereplication did not occur in SIC1ΔP HA rad53Δ sml1-1 cells, regardless of DMC1 status ( Figure S6), even after 48 hr (data not shown). As an alternative genetic test, we turned to RAD9, which encodes a protein that mediates Rad53 activation in many circumstances (Sun et al. 1998;Vialard et al. 1998;Gilbert et al. 2001). By contrast to rad53D, rad9Δ did not prevent DNA rereplication in SIC1ΔP HA cells ( Figure S6). Importantly, RAD9 was not required for suppressing DNA rereplication in SIC1ΔP HA dmc1D cells ( Figure 6A), indicating that RAD9 was not involved in this checkpoint response.
Because our results precluded analysis of Rad53 function through gene deletion, we explored the possibility of a Rad53, or Rad53-like, function through different means. In response to replication fork stalling or DNA damage during S phase of the mitotic cell cycle, Rad53 is activated to catalyze phosphorylation of Dbf4 and Sld3, thereby preventing firing of late origins (Lopez-Mosqueda et al. 2010;Zegerman and Diffley 2010). We examined two different types of mutants to determine whether this process could be involved in our system. The first involved cells containing mutant alleles of both DBF4 and SLD3, encoding proteins with alterations in important Rad53-targeted phosphorylation sites. SIC1ΔP HA dmc1Δ cells containing the dbf4-4A and sld3-38A alleles exhibited DNA rereplication, indicating that phosphorylation of Dbf4 or Sld3, or both proteins, was required for full prevention of DNA rereplication (Figure 7 and Figure S1). We noticed that, as in the case of SIC1ΔP HA dmc1Δ h2A-S129A cells, these cells appeared to rereplicate their DNA less than SIC1ΔP HA cells, in that they did not exhibit .8C DNA content. In SIC1ΔP HA dbf4-4A sld3-38A cells, we did observe cells with .8C DNA content, indicating that the phosphorylation site mutations themselves were not responsible for limiting DNA rereplication ( Figure S7). The second type of mutant cells contained the sld3-38A allele and mcm5-bob1, a mutant allele that bypasses the essential function of DBF4 (Hardy et al. 1997). DNA rereplication was not observed in SIC1ΔP HA dmc1Δ mcm5-bob1 sld3-38A cells (Figure 7). We confirmed that the mcm5-bob1 allele did not prevent DNA rereplication in our system ( Figure S7). While it might be expected that cells with altered Dbf4 phosphorylation sites would behave similarly to cells with mcm5-bob1, as in the mitotic cell cycle studies, certain experimental factors may account for this discrepancy (see Discussion). These data suggest that Dbf4 phosphorylation was sufficient to prevent DNA rereplication, at least in the context of the mcm5-bob1 allele. It is noted that either Dbf4 or Sld3 phosphorylation alone can contribute to prevention of late origin firing in mitotic S phase (Lopez-Mosqueda et al. 2010;Zegerman and Diffley 2010). We conclude that sites normally phosphorylated by Rad53 in the mitotic cell cycle also functioned to prevent DNA replication during meiosis under our conditions that promote DNA rereplication.
To further investigate a possible Rad53 function, we examined Rad53 enzymatic activity. While Rad53 isoforms with reduced electrophoretic mobility are indicative of its phosphorylation and activation, a more sensitive method is to analyze Rad53 autophosphorylation in situ (Pellicioli et al. 1999). As can be seen in Figure 6B, a slight Rad53 activation was observed at the late time point in the SIC1DP HA , SIC1DP HA dmc1D, and SIC1DP HA dmc1D rad9D strains. This activation was negligible relative to HU-induced Rad53 activation in mitotic cells. While it appeared that the Rad53 activation was a bit higher in the SIC1DP dmc1D cells, it is unlikely that this degree of activation could have been responsible for the checkpoint, given that heightened activation was not observed in the checkpoint-proficient SIC1DP HA dmc1D rad9D cells. It is possible that persistence of DSBs and accumulation of DNA damage over time in SIC1DP HA dmc1D cells led to low level Rad9-dependent Rad53 phosphorylation.

DISCUSSION
When S. cerevisiae cells deleted for DMC1 are induced to enter the meiotic program, a response is activated that inhibits recombination between sister chromatids and prevents progression into the meiotic divisions (Bishop et al. 1992;Xu et al. 1997;Wan et al. 2004;Niu et al. 2005). We have observed that cells undergoing meiosis but engineered to rereplicate their DNA through expression of SIC1ΔP HA also respond to deletion of DMC1 by preventing extra DNA replication (Sawarynski et al. 2009). Our previous studies indicated that this response requires RAD17, suggesting a checkpoint mechanism. Our further genetic analysis shown here has confirmed that a checkpoint response is involved.
We observed differences in the degree of DNA rereplication recovery in SIC1ΔP HA dmc1D cells depending on which additional gene was deleted or mutated. While experimental variability makes it difficult to be conclusive about these differences, we observed trends in particular circumstances. For example, removal of genes encoding certain proteins, such as Mek1, that serve to prevent both intersister repair and meiotic progression led to extensive DNA rereplication including cells containing .8C DNA content. Our experiments further suggest that the MEK1 function to prevent intersister repair, and to thereby retain the checkpoint signal originating from unrepaired DSBs, is operative in preventing DNA rereplication in our cells. By contrast to these gene deletions, mutation of genes to abrogate H2A or Dbf4 and Sld3 phosphorylation led to less DNA rereplication, with few cells detected containing .8C DNA content. This difference might indicate that the degree of DNA rereplication is limited by the presence of unrepaired and resected DSBs, although we did observe considerable g-H2A staining in cells, regardless of checkpoint status. In addition, more than a single checkpoint mechanism may be involved in preventing DNA rereplication, with only one being absent in certain cases such as SIC1ΔP HA dmc1D dbf4-4A sld3-38A cells. In this sense, perhaps the checkpoint pathway that we have uncovered affects only a subset of origins. Regardless, the data provide clear evidence that the meiotic recombination checkpoint can target the DNA replication machinery.
The meiotic recombination checkpoint response that prevents DNA rereplication shares several components with cell cycle DNA damage response checkpoint mechanisms. Genes encoding Mec1 and Rad17, and, presumably, the entire 9-1-1 complex that includes Rad17 and facilitates Mec1 function, were required in our system. It is interesting to note that DNA rereplication itself in the mitotic cell cycle induces a checkpoint response dependent on MEC1 and RAD17 that restricts the extent of DNA rereplication (Archambault et al. 2005). In our case, we observed slightly less DNA rereplication upon deletion of MEC1 and SML1 in SIC1ΔP HA dmc1D cells when compared with SIC1ΔP HA cells deleted for genes such as MEK1, particularly with regard to .8C DNA content cells. (The possibility is noted that deletion of MEC1 and SML1 might have had a minor effect on DNA rereplication in SIC1ΔP HA cells as well). It has been reported that dmc1D mec1-1 cells continue to progress through meiosis with unrepaired DSBs (Lydall et al. 1996); as suggested above, the presence of unrepaired DSBs may influence the degree of DNA rereplication in our system. We did not interrogate TEL1, which encodes a close relative of Mec1 that is involved in DNA damage response pathways, including the meiotic recombination checkpoint response (Greenwell et al. 1995;Morrow et al. 1995;Usui et al. 2001). Like Mec1, Tel1 catalyzes Hop1 phosphorylation, which is required for Mek1 activation (Carballo et al. 2008). It is possible that the phenotypic difference between SIC1ΔP HA dmc1D mec1D sml1D and SIC1ΔP HA dmc1D mek1D cells (observed despite the fact that Mek1 functions downstream of Mec1) is due to the presence of Tel1. In addition, there might exist downstream effectors of Mec1, other than Mek1, that have an influence by promoting DNA rereplication.
Mec1 activation involves the interaction of its partner Ddc2 with single-stranded-DNA bound replication protein A (Zou and Elledge 2003), which, in the case of dmc1D cells would be formed readily due to the highly resected Spo11-generated DSBs (Bishop et al. 1992). In turn, Mec1 can catalyze formation of g-H2A, which we found contributed to prevention of DNA rereplication, as did DOT1, encoding the enzyme that generates H3meK79. Because g-H2A was abundant in SIC1ΔP HA cells that underwent DNA rereplication, and H3meK79 was likely to be abundant as well (van Leeuwen et al. 2002), these two histone modifications are necessary for the full checkpoint response, but not sufficient. In the mitotic cell cycle, particularly with respect to the G1 DNA damage checkpoint response, these modifications (as well as Rad6-Bre1 mediated histone H2B ubiquitylation required for H3meK79 generation) are important for Rad9 recruitment and Rad53 activation (Giannattasio et al. 2005;Javaheri et al. 2006;Hammet et al. 2007). In the checkpoint response that prevents DNA rereplication, neither RAD9 nor RAD53 appeared to be involved. While it is theoretically possible that Rad53 can be activated by a mechanism that cannot be detected by conventional means (electrophoretic mobility shift or in situ autophosphorylation), this possibility seems remote. In fact, our results are consistent with the fact that the Rad9-Rad53 axis is not involved in the meiotic recombination checkpoint that serves to prevent pachytene exit and progression through the meiotic divisions (Bishop et al. 1992;Lydall et al. 1996;Bailis and Roeder 2000). Of particular interest, however, is that residues in Dbf4, and presumably Sld3 as well, that are phosphorylated through Rad53 catalysis in the mitotic cell cycle to prevent late origin firing (Lopez-Mosqueda et al. 2010;Zegerman and Diffley 2010), also function in the meiotic response to prevent DNA rereplication. By contrast to the mitotic cell cycle observation (Zegerman and Diffley 2010), we found that the mcm5-bob1 allele, which eliminates the requirement for the Dbf4-Cdc7 protein kinase in DNA replication initiation (Hardy et al. 1997), did not substitute for the dbf4-4A mutant allele as it did in the mitotic cell cycle studies. This distinction could reflect intrinsic differences in DNA replication regulation during the mitotic cell cycle and meiosis; alternatively, it could be due to the lowered CDK activity in our engineered cells.
Taken together, our results suggest a response in which a protein kinase that recognizes motifs in a similar manner to Rad53 is activated through a mechanism equivalent in many ways to the one that operates in response to DNA damage during the mitotic cell cycle (see Figure 8). One likely candidate for this kinase activity would be Mek1, given its structural similarity to Rad53 and its meiosis-specific expression (Rockmill and Roeder 1991;Leem and Ogawa 1992;Bailis and Roeder 2000). However, our data argue against this possibility because Figure 8 A meiotic recombination checkpoint response can inhibit DNA rereplication. This diagram based on our genetic analysis depicts certain key protein components in a pathway that leads from accumulation of unrepaired DSBs to inhibition of DNA rereplication in the SIC1ΔP HA system. Also shown is an outline of the pathways that prevent intersister repair and meiotic progression in the normal checkpoint. See text for details. SIC1ΔP HA dmc1D cells devoid of both MEK1 and RAD54, designed to examine MEK1 checkpoint function specifically, did not display DNA rereplication. Furthermore, while a peptide-based investigation into the phosphorylation site specificity of yeast protein kinases has placed Rad53 and Mek1 into the same largest group of five clusterings, the two kinases are not closely related in this regard, and exhibit a considerable difference in their degree of specificity (Mok et al. 2010). Another kinase with some physical similarity to Rad53 is Dun1, which, in response to genotoxic stress, operates downstream of Rad53 to regulate nucleotide pool levels and transcription (Allen et al. 1994;Zhao and Rothstein 2002). However, biochemical studies suggest that the two enzymes have different specificities (Zheng et al. 1993;Sanchez et al. 1997;Sidorova and Breeden 2003;Uchiki et al. 2004;Chen et al. 2007;Mok et al. 2010), and a comprehensive analysis of transcriptional regulation upon DNA damage indicates different targeting by Rad53 and Dun1 (Jaehnig et al. 2013). Therefore, evidence does not exist to indicate that Mek1 or Dun1 would likely catalyze phosphorylation of the same sites in Dbf4 and Sld3 as Rad53 does, suggesting that a different Rad53-like enzyme is present in meiosis that can influence initiation of DNA replication.